Spatial dissimilarity analysis in single cell transcriptomics

Abstract

The spatial dissimilarity method is a new statistical tool to uncover complex bivariate relationships in single cells and spatial transcriptomics data, thereby addressing unresolved issues such as alternative splicing and allele-specific gene expression. By applying our spatial dissimilarity analysis method on datasets of neurons, tumor, and normal cells, we identified thousands of alternatively spliced genes and instances of allele-specific gene expression. This is particularly evident in the analysis of quiescent and transitioning cell states, where our method reveals the nuanced gene expression dynamics associated with these states. Notably, our findings highlight how allele-specific genetic variants can provide insights into the subclone architecture of normal cells and cancer cells, offering a more comprehensive understanding of cellular heterogeneity and a new insight to understand gene function at specific cell types. The spatial dissimilarity analysis method deepens our comprehension of cellular complexity and gene expression dynamics during cell state transitions.